Need Some Dimensional Info

[bboyd]

Here's the problem let's say you have a dim of 1.00 tolerance of +/- .001 and you use a gauge that meets the factor of 10 on the tolerance ( accuracy of .0002) can I accept the part if it reads 1.0011 or .9989. or even 1.0014 based on the tolerance being only 3 decimal places? Is there a spec or std that proves this.

Thanks

[cmmguy]

Some diehards will say yes and the reason is that the "real value" might actually be 1.0013 due to the error. Some will contend that you can round the number to the nearest decimal place(3). But the problem with the latter is that the dimension should probably have said 1.0000 +/-.001 because of the tolerance. You shouldn't have a tolerance of plus or minus 1 unit.

As an inspector - reject it.

[bboyd]

That is exactly the issue, I only have 3 decimal places but we are rejecting the part based on 4 decimal places. Can I round the result down if it is out .0004 or less? I have been told this is in a spec somewhere but I have never seen it.

[cmmguy]

I dont see how you can round it. The 3 decimal places give you only +/- one unit. You would have to go to four in order to get the 10 to one rule that you indicated that you wanted. By rounding it, you are ignoring the 10 to one rule.

.0001" represents 5% of your tolerance.

[John Kingston]

ASTM E 29 is Standard Practice for Using Significant Digits in Test Data to Determine Conformance with Specifications
Rounding procedures are covered in ASTM Manual 7.
If you have the CMI Primer from ASQ, it covers rounding in the Inspection and Testing section.( Actually it quotes ASTM Manual 7)

[Wes Cisco]

I was always taught that the limit is the limit. Either you are in tol or not. So even if your tol is +/- .5 if you are .0001 over the high you are out of tol. If you want to take into account measurement uncertainty then you only use it to narrow the tol of what you will accept, but never to enlarge it. So if your tol is +/-.010 and your uncertainty is .0004, then you could narrow your tol to +/- .0096 but you could not grow it to +/-.0104. Of course I have also heard the following phrases often: "They won't check it, ship it.", "Let them prove it is out.", "What is a few tenths amoung friends?", etc.

[cmmguy]

ASTM E 29 is Standard Practice for Using Significant Digits in Test Data to Determine Conformance with Specifications
Rounding procedures are covered in ASTM Manual 7.
If you have the CMI Primer from ASQ, it covers rounding in the Inspection and Testing section.( Actually it quotes ASTM Manual 7)

And the answer is....?

... If you want to take into account measurement uncertainty then you only use it to narrow the tol of what you will accept, but never to enlarge it. So if your tol is +/-.010 and your uncertainty is .0004, then you could narrow your tol to +/- .0096 but you could not grow it to +/-.0104

Thank you, that is what I was trying to say earlier but you put it a much better way. Gage uncertainty always goes against you, never for you.

[Jan d.]

If you want to take into account measurement uncertainty then you only use it to narrow the tol of what you will accept, but never to enlarge it. So if your tol is +/-.010 and your uncertainty is .0004, then you could narrow your tol to +/- .0096 but you could not grow it to +/-.0104.

I agree with you Wes. We have the same discussion here all the time. The problem is how to determine the uncertainty. It has little to do with accuracy and resolution.

When we have close tolerance measurements, we try to subtract the uncertainty from the tolerance band. Like you argue, we would us the +/-0.0096. That would guarantee us that the measurement is always within the print limits. But how do you know that the uncertainty is 0.0004"????

Determining the uncertainty of the measurement can be hard. You have to look at your complete measuring plan (like how many PH10 angle rotations do you have), not just the machine.



Jan.

[cmmguy]

I agree with you Wes. We have the same discussion here all the time. The problem is how to determine the uncertainty. It has little to do with accuracy and resolution.

When we have close tolerance measurements, we try to subtract the uncertainty from the tolerance band. Like you argue, we would us the +/-0.0096. That would guarantee us that the measurement is always within the print limits. But how do you know that the uncertainty is 0.0004"????

Determining the uncertainty of the measurement can be hard. You have to look at your complete measuring plan (like how many PH10 angle rotations do you have), not just the machine.



Jan.

The calibration house should provide you with the uncertainty values for the gage. The GR&R evaluates the other parameters.

[kingsld1]

In Y14.5 paragraph 2.4 (page 25) says " All limits are absolute. Dimensional limits, regardless of the number of decimal places, are used as if they were continued with zeros......"

[John Kingston]

[QUOTE=cmmguy;37700]And the answer is....?[QUOTE]

I didn't give a complete answer because every company uses its' own standard.
Are you allowed to round? Depends on the customer.
If you round away the .0002" will the part still function?
So many questions to ask. I just wanted to lead him to the specs. and let him (and his organization make a decision.)

[Andrew Gardner]

Of course I have also heard the following phrases often: "They won't check it, ship it.", "Let them prove it is out.", "What is a few tenths amoung friends?", etc.

When in doubt, ship it out!

[Farmer]

If in doubt, chuck it out )

[bboyd]

In Y14.5 paragraph 2.4 (page 25) says " All limits are absolute. Dimensional limits, regardless of the number of decimal places, are used as if they were continued with zeros......"

Thanks for all the info this is exactly what I needed for backup of my original position.